Language Models are Super Mario: Absorbing Abilities from Homologous Models as a Free Lunch
Paper
• 2311.03099 • Published
• 30
This is a merge of pre-trained language models created using mergekit.
This model was merged using the DARE TIES merge method using google/gemma-2b-it as a base.
The following models were included in the merge:
The following YAML configuration was used to produce this model:
merge_method: dare_ties
models:
- model: rhaymison/gemma-portuguese-luana-2b
parameters:
weight: 0.5
density: 0.4
- model: google/gemma-2b-it
parameters:
weight: 0.5
density: 0.4
parameters:
int8_mask: true
dtype: bfloat16
tokenizer:
source: union
base_model: google/gemma-2b-it
write_readme: README.md